DETAILED ACTION
Remarks
Claims 1, 3-11 and 13-22 have been examined and rejected. This Office Action is responsive to the continued examination request.
Notice of Pre-AIA or AIA Status
The present application, filed on or after March 16, 2013, is being examined under the first inventor to file provisions of the AIA .
Claims 1, 3-11 and 13-22 are presented for examination.
Respond to Amendment
The amendment filed 03/04/2026 has been entered. Claims 1, 3, 11, 13 and 20-22 have been amended. Claims 1, 3-11 and 13-22 are pending in the application.
Claim Rejections - 35 USC § 103
The following is a quotation of 35 U.S.C. 103 which forms the basis for all obviousness rejections set forth in this Office action:
A patent for a claimed invention may not be obtained, notwithstanding that the claimed invention is not identically disclosed as set forth in section 102, if the differences between the claimed invention and the prior art are such that the claimed invention as a whole would have been obvious before the effective filing date of the claimed invention to a person having ordinary skill in the art to which the claimed invention pertains. Patentability shall not be negated by the manner in which the invention was made.
Claims 1, 3-8 and 10, 11, 13-22 are rejected under 35 U.S.C. 103 as being unpatentable over Benoit et al (“Quantization and Training of Neural Networks for Efficient Integer-Arithmetic-Only Inference”) hereafter Benoit, further in view of Ruihao et al (“Differentiable Soft Quantization: Bridging Full-Precision and Low-Bit Neural Networks”) hereafter Ruihao, further in view of Nagel et al (US 20200302299 A1) hereafter Nagel, and further in view of Sather et al (US 12136039 B1) hereafter Sather.
Benoit was cited in the IDS filed on 03/22/2022.
Ruihao was cited in the IDS filed on 03/22/2022.
With respect to claim 1, Benoit teaches a computer-implemented method of training a neural network that comprises a plurality of computational blocks, the neural network having been previously configured for an inference task and including a set of real-valued weights, the method (the method involves integer-arithmetic-only inference of a convolution layer and training with simulated quantization of the convolution layer. Frameworks such as quantization scheme, quantized inference framework and quantized training framework those may be applied for efficient classification and detection systems based on MobileNets and significant improvements in latency-vs-accuracy tradeoffs in ImageNet classification, object detection and other inference tasks [Figure 1.1. Integer-arithmetic-only quantization]), comprising:
performing a plurality of training iterations, each training iteration comprising:
for each computational block, applying a respective quantization function to the set of respective real-valued weights of the computational block to generate a respective set of quantized weights that are scaled based on a respective scaling factor to fall within a respective quantization range that is symmetrically centered at zero and comprises a defined number of uniform quantization levels corresponding to integer multiples of the respective scaling factor (the training includes floating point training and quantizing the resulting weights. Weights are quantized before they are convolved with the input. For each layer, quantization is performed by a number of quantization levels and range, where r is real-valued weight, [a;b] is the quantization range, n is quantization levels. Quantization ranges are treated differently, the quantized weights are ranged in [-127, 127], wherein the center is 0. The learned quantization parameters map to the scale S and zero-point Z in equation S = s(a,b,n) and Z = z(a,b,n). The scaling factor such as quantization scales S1, S2, S3 are taken into consideration when computing quantized weights [pages 4, 5; 2.4. Implementation of a typical fused layer, 3. Training with simulated quantization]);
for each computational block, computing a set of respective output activations for the computational block based on a respective set of input activations and the respective set of quantized weights (Computing the quantized bias-addition includes computing the product of the weights and the input activations based on the quantization scales. The output activations are computed based on the quantized weights and the input activations, wherein the three steps in the training process includes scale down to final scale used by 8-bit output activations, cast down to unit8 and apply the activation function to yield the final 8-bit output activation [page 4, 2.4. Implementation of a typical fused layer]); and
when performing the plurality of training iterations, incrementally reducing a smoothness of the respective quantization functions applied by the computational blocks for multiple training iterations of the plurality of training iterations (for activations, ranges depend on the input. To estimate the ranges, we collect [a;b] ranges seen on activations during training and then aggregate them via exponential moving averages with the smoothing parameter being close to 1, so that observed ranges are smoothed across thousands of training steps. This concept may be applied to the quantization function for multiple training steps [page 5, 3.1. Learning quantization ranges]); and
storing, for each of the computational blocks, a quantized weights version of the adjusted set of respective real-valued weights that optimizes performance of the neural network on the inference task, at a completion of the plurality of training iterations (one approach of simulated quantization is to train in floating point and then quantize the resulting weights, sometimes with additional post-quantization training for fine-tuning. Some post-quantization include: large differences in ranges of weights for different output channels, and outlier weight values that make all remaining weights less precise after quantization. The forward propagation pass simulates quantized inference as it will happen in the inference engine by implementing in floating-point arithmetic the rounding behavior of the quantization scheme. The system of Benoit represents the preparation (simulation) and the finalization (storage) of a model specifically designed for high-performance inference tasks. The “simulated quantized inference” during training is the mechanism used to achieve the “optimized performance” of the final “quantized weights.” [page 5, 3. Training with simulated quantization]).
However, Benoit does not disclose computing a cost for the training iteration based on the respective output activations of the computational blocks and relative alignments of the respective quantized weights of the computational blocks with the uniform quantization levels of the respective quantization ranges, the cost including a prediction task error and a regularization cost, the regularization cost including a weight regularization term, and a scaling factor regularization term that enforces weight-to-quantization-level alignment, wherein computing the regularization cost comprises applying the scaling factor regularization function to output regularization cost values based on the respective quantized weights and the respective scaling factors, the scaling factor regularization function being configured to generate a regularization cost value that decreases the closer that the respective quantized weights each align with one of the uniform quantization levels; and for each computational block, adjusting the set of respective real- valued weights and the respective scaling factor with an objective of reducing the computed cost in one or more following training iterations.
In the same field of endeavor, Ruihao teaches computing a cost for the training iteration based on the respective output activations of the computational blocks and relative alignments of the respective quantized weights of the computational blocks with the uniform quantization levels of the respective quantization ranges (the differentiable soft quantization (DSQ) function is proposed to optimize the DSQ training and network parameters. The algorithm 1 of the training includes input activation a, parameters weight w and similarity factor, and output activation o. The computational cost of SADDW and MLA are quite different, SADDW takes extra computational cost than MLA. DSQ function is used to approximately model the uniform quantizer, wherein k is the coefficient that determines the shape of the function, and k is aligned with the quantization levels [page 3, 3.2. Quantization Function; page 5, 3.5. Training and Deploying and Figure 4]); wherein computing the regularization cost comprises applying the scaling factor regularization function to output regularization cost values based on the respective quantized weights and the respective scaling factors, the scaling factor regularization function being configured to generate a regularization cost value that decreases the closer that the respective quantized weights each align with one of the uniform quantization levels ([the differentiable soft quantization (DSQ) function is proposed to optimize the DSQ training and network parameters. The algorithm 1 of the training includes input activation a, parameters weight w and similarity factor, and output activation o. The computational cost of SADDW and MLA are quite different, SADDW takes extra computational cost than MLA. DSQ function is used to approximately model the uniform quantizer, wherein k is the coefficient that determines the shape of the function, and k is aligned with the quantization levels [page 3, 3.2. Quantization Function; page 5, 3.5. Training and Deploying and Figure 4]]);
It would have been obvious to one of ordinary skill in the art before the effective filing date of the invention to have incorporated the concept of applying differentiable soft quantization to bridge the gap between full-precision and low-bit networks as suggested by Ruihao into the concept of applying quantization scheme technique using integer-only arithmetic as suggested by Benoit because both of the systems addressing the process of applying the quantization process on the weights and the input activations to get the output activations based on the scaling factor. Doing so would be desirable because the concept of Benoit would be more efficient by reducing the computational cost and expensive memory by employing a more promising network compression solution such as DSQ to reduce the network storage and accelerate the inference speed using different types of quantizers (Ruihao, [page 1, 1. Introduction]).
However, the combination of Benoit and Ruihao does not teach the cost including a prediction task error and a regularization cost, the regularization cost including a weight regularization term, and a scaling factor regularization term that enforces weight-to-quantization-level alignment; and for each computational block, adjusting the set of respective real- valued weights and the respective scaling factor with an objective of reducing the computed cost in one or more following training iterations.
In the same field of endeavor, Nagel teaches for each computational block, adjusting the set of respective real- valued weights and the respective scaling factor with an objective of reducing the computed cost in one or more following training iterations (a neural network quantization method for performing cross layer is illustrated. The processor may scale each of the output channel weights by corresponding scaling factor. The processor may change or adjust the weights within the layer by scaling the output channel weights by a corresponding scaling factor [par. 0063-0065 and FIG. 4]).
It would have been obvious to one of ordinary skill in the art before the effective filing date of the invention to have incorporated the concept of scaling each output channel weight of a first layer of the neural network by a corresponding scaling factor as suggested by Nagel into the combination of Benoit and Ruihao because all of the systems addressing the process of performing quantization in the neural network. Doing so would be desirable because the combination of Benoit and Ruihao would be more efficient by scaling each weight of each layer of the neural work by a corresponding scaling factor, as well as determining the scaling factor to equalize ranges of weight tensors or output channel weights to minimize the quantization error (Nagel, [par. 0003-0004]).
However, the combination of Benoit, Ruihao and Nagel does not explicitly disclose the cost including a prediction task error and a regularization cost, the regularization cost including a weight regularization term, and a scaling factor regularization term that enforces weight-to-quantization-level alignment.
In the same field of endeavor, Sather teaches the cost including a prediction task error and a regularization cost, the regularization cost including a weight regularization term, and a scaling factor regularization term that enforces weight-to-quantization-level alignment (a training of a neural network involves in defining a loss function for the network that measures an error (loss) of an actual output of the network for a particular input compared to a pre-defined expected (ground truth) output for that particular input. A constraint generator uses the loss matrix terms to generate the loss-aware penalty term for use by the error calculator and propagator. A Hessian matrix function is used to calculate the quantized weights. A parameter of the loss-aware function controls the regularization forces that dictates how fast the quantization takes place. Scaling factor regularization function is defined that push the weights of computational block to respective quantized values that are a closest multiple of scaling factor. Initializing the quantized weight values in order to train the quantized network requires determining the scaling factor for each layer and initializing each weight to one of the scaled ternary values. The weight values in the k-th layer are initialized to the quantized values by assigning each weight value to the nearest scaled ternary value. Prediction error can be a function that computes an error value that represents a difference between predictions output by NN and target values. The training process computes error of the predicted outputs versus the expected outputs for the selected inputs. The loss function quantifies the difference between the predicted output and the actual output for each input [col. 16, lines 30-45; col. 23, lines 5-45; col. 28, line 55 – col. 29, line 5; col. 35, lines 25-40]).
It would have been obvious to one of ordinary skill in the art before the effective filing date of the invention to have incorporated the concept of training multiple parameters of a machine-trained network subject to a sparsity constraint as suggested by Sather into the combination of Benoit, Ruihao and Nagel because all of the systems addressing the process of performing quantization in the neural network. Doing so would be desirable because the combination of Benoit, Ruihao and Nagel would be more efficient by initializing the weight values to the quantized values by assigning each weight value to the nearest scaled ternary value and setting a predefined portion of the weight values to a value of zero as part of or immediately after initialization in order to minimize the impact on the loss (Sather, [col. 28, line 55 – col. 29, line 5]).
With respect to claim 3, the combination of Benoit, Ruihao, Nagel and Sather teaches wherein the neural network comprises an input block prior to the plurality of computational blocks, and an output block following the plurality of computational blocks, the input block, plurality of computational blocks, and output block arranged as respective layers of the neural network to collectively process input feature tensors received at the input block representing objects and output, from the output block, respective predictions for the objects (Benoit, input block, output block and a plurality of computational blocks are described in the training of integer-arithmetic-only inference and training with simulated quantization. All input variables and computations are carried out using 32-bit floating-point arithmetic [page 2, 1. Introduction and Figure 1.1]), and
wherein, for each training iteration, the respective set of input activations for each of the plurality of computational blocks following a first computational block is the set of output activations computed by a preceding computation block of the plurality of computational blocks, and each training iteration comprises, for each computational block, applying a respective activation quantization function to the respective set of respective set of input activations to generate a respective set of quantized activations (Benoit, for each layer, the quantization function is applied with a number of quantization levels and quantization range, with r is real-valued number to be quantized and n is quantization level which is fixed for all layers [page 5, 3. Training with simulated quantization]), wherein for each computational block, computing the set of respective output activations for the computational block is based on a matrix multiplication of the respective set of quantized activations and the respective set of quantized weights for the computational block (Benoit, the real numbers are applied in the multiplication of two square NxN matrices along with the quantization parameters to perform the matrix multiplication. The quantization scales S1, S2, S3 are also used in the computation to generate the output activations [page 3, 2.2. Integer-arithmetic-only matrix multiplication]);
wherein, for each training iteration, computing the cost comprises computing the prediction task error as an error between the respective predictions for the objects and expected values for the objects (Ruihao, clipping and rounding together cause the quantization error. When the quantizer clips more, the clipping error increases and the rounding error decreases. When clipping error dominates the whole quantization error, the outlier’s gradients will be large and thus serve as the major power for weight updating [page 4, 3.4. Balancing clipping error and rounding error]).
With respect to claim 4, the combination of Benoit, Ruihao, Nagel and Sather teaches wherein, for each computational block, applying the respective activation quantization function generates the respective set of quantized activations scaled within a respective activation quantization range that is symmetrically centered at zero and comprises a defined number of uniform activation quantization levels (Ruihao, the differentiable soft quantization (DSQ) function is proposed to optimize the DSQ training and network parameters. The algorithm 1 of the training includes input activation a, parameters weight w and similarity factor, and output activation o. The computational cost of SADDW and MLA are quite different, SADDW takes extra computational cost than MLA. DSQ function is used to approximately model the uniform quantizer, wherein k is the coefficient that determines the shape of the function, and k is aligned with the quantization levels [page 3, 3.2. Quantization Function; page 5, 3.5. Training and Deploying and Figure 4]).
With respect to claim 5, the combination of Benoit, Ruihao, Nagel and Sather teaches wherein the computational blocks include at least one computational block that implements one of a fully connected neural network layer or a convolution neural network layer (Benoit, activations are quantized at points where they would be during inference after the activation function is applied to a convolutional or fully connected layer’s output [page 5, 3. Training with simulated quantization]).
With respect to claim 6, the combination of Benoit, Ruihao, Nagel and Sather teaches wherein for each computational block, the respective quantization function is a piecewise function comprising a plurality of repeated, shifted functions that each correspond to a respective uniform quantization level, and wherein incrementally reducing the smoothness of the respective quantization functions comprises incrementally increasing a slope of the function (Sather, the optimization method for minimizing the output loss function iteratively repeated over multiple epochs is referred to the stochastic gradient descent. The weight modifier uses gradients to modify the weight values for the next training iteration. The process is repeated for multiple minibatches of inputs until the weight values have converged to optimized floating-point values [col. 15, lines 30-45; col. 20, lines 30-55]; and Benoit, for activations, ranges depend on the input. To estimate the ranges, we collect [a;b] ranges seen on activations during training and then aggregate them via exponential moving averages with the smoothing parameter being close to 1, so that observed ranges are smoothed across thousands of training steps. This concept may be applied to the quantization function for multiple training steps [page 5, 3.1. Learning quantization ranges]).
With respect to claim 7, the combination of Benoit, Ruihao, Nagel and Sather teaches wherein adjusting the set of respective real-valued weights and the respective scaling factor for each computational block is performed using a derivative of a corresponding one of the plurality of repeated, shifted functions for at least some of the plurality of training iterations (Sather, the diagonal values of Hessian matrix are second-order partial derivatives of the loss function with respect to each weight in the MT network. The loss increase for a particular weight is linearly related to a corresponding diagonal term of Hessian matrix. The Fisher approximation method estimates the Hessian by computing first-order derivatives of the loss function [col. 3, lines 40-65; col. 17, lines 10-40; col. 25, line 60 – col. 26, line 40]).
With respect to claim 8, the combination of Benoit, Ruihao, Nagel and Sather teaches wherein incrementally reducing the smoothness of the respective quantization functions is performed in a linear manner across at least a first group of the plurality of training iterations and is suspended when a predetermined criteria is reached, following which a quantization function of constant smoothness is used as the respective quantization functions for a remainder of the plurality of training iterations (Sather, the training process adjusts the weight values based on the relative rates of change and a training rate factor. After the weights are updated, the process determines whether to perform additional training. A minimization process (stochastic gradient descent minimizer) is used to determine when to stop training the network. The training system only stops training the network once the weights have changed by less than a threshold for a particular number for training iterations. The trained weights will again be projected onto their quantized values and sparsified, only a certain number of training iterations are performed before performing the projection/sparsification process [col. 36, lines 10-55]).
With respect to claim 10, the combination of Benoit, Ruihao, Nagel and Sather teaches deploying a trained version of the neural network that includes the quantized weights version for each of the computational blocks, the trained version of the neural network representinq a compressed neural network (DSQ is implemented using Pytorch as a flexible module that can be easily inserted to the binary or uniform quantization models. Two strategies are used: moving average statistics and optimization by backward propagation. ARM NEON 8-bit with GEMM kernels are also used that accelerate the inference even when using the extreme lower bits [pages 5-8, 4. Experiments]).
With respect to claim 11, it is a processing unit that is corresponding to the method of claim 1. Therefore, it is rejected for the same reason as claimed in claim 1 above.
With respect to claim 12, it is a processing unit that is corresponding to the method of claim 2. Therefore, it is rejected for the same reason as claimed in claim 2 above.
With respect to claim 13, it is a processing unit that is corresponding to the method of claim 3. Therefore, it is rejected for the same reason as claimed in claim 3 above.
With respect to claim 14, it is a processing unit that is corresponding to the method of claim 4. Therefore, it is rejected for the same reason as claimed in claim 4 above.
With respect to claim 15, it is a processing unit that is corresponding to the method of claim 5. Therefore, it is rejected for the same reason as claimed in claim 5 above.
With respect to claim 16, it is a processing unit that is corresponding to the method of claim 6. Therefore, it is rejected for the same reason as claimed in claim 6 above.
With respect to claim 17, it is a processing unit that is corresponding to the method of claim 7. Therefore, it is rejected for the same reason as claimed in claim 7 above.
With respect to claim 18, it is a processing unit that is corresponding to the method of claim 8. Therefore, it is rejected for the same reason as claimed in claim 8 above.
With respect to claim 19, it is a processing unit that is corresponding to the method of claim 10. Therefore, it is rejected for the same reason as claimed in claim 10 above.
With respect to claim 20, it is a non-transitory computer readable storage that is corresponding to the method of claim 1. Therefore, it is rejected for the same reason as claimed in claim 1 above.
With respect to claim 21, the combination of Benoit, Ruihao, Nagel and Sather teaches deploying a trained version of the neural network that includes the quantized weights version for each of the computational blocks on a computationally constrained hardware device, wherein the quantized weights version of the adjusted set of respective real-valued weights optimize performance of the neural network on the inference task when the neural network is deployed on the computationally constrained hardware device (For deploying on devices with limited computing resources, low-bit computation kernels are implemented to accelerate the inference on ARM architecture. In convolution networks, multiply and accumulation are the core operations of General Matrix Multiply which can be efficiently completed by the MLA instruction on ARM NEON. DSQ is implemented using Pytorch as a flexible module that can be easily inserted to the binary or uniform quantization models. Two strategies are used: moving average statistics and optimization by backward propagation [pages 5-8, 3.5. Training and Deploying & 4. Experiments]).
With respect to claim 22, it is a processing unit that is corresponding to the method of claim 21. Therefore, it is rejected for the same reason as claimed in claim 21 above.
Claim 9 is rejected under 35 U.S.C. 103 as being unpatentable over Benoit et al (“Quantization and Training of Neural Networks for Efficient Integer-Arithmetic-Only Inference”) hereafter Benoit, further in view of Ruihao et al (“Differentiable Soft Quantization: Bridging Full-Precision and Low-Bit Neural Networks”) hereafter Ruihao, further in view of Nagel et al (US 20200302299 A1) hereafter Nagel, further in view of Sather et al (US 12136039 B1) hereafter Sather, as claimed in claim 1 above, and further in view of Bialkowski et al (US 20100008594 A1) hereafter Bialkowski.
Benoit was cited in the IDS filed on 03/22/2022.
Ruihao was cited in the IDS filed on 03/22/2022.
With respect to claim 9, the combination of Benoit, Ruihao, Nagel and Sather teaches all limitations as claimed in claim 1 above.
However, the combination of Benoit, Ruihao, Nagel and Sather does not disclose wherein the defined number of uniform quantization levels is 15.
In the same field of endeavor, Bialkowski teaches wherein the defined number of uniform quantization levels is 15 (an image quality can be seen using quantization level using one or two quantization when used in a video coding method. With a quantization level of 15 the amplitudes from 0-14 or from 15-29 are each summarized to a reconstruction value [par. 0008]).
It would have been obvious to one of ordinary skill in the art before the effective filing date of the invention to have incorporated the concept of transcoding the quantized digital signals as suggested by Bialkowski into the combination of Benoit, Ruihao, Nagel and Sather because all of the systems addressing the process of performing quantization in the neural network. Doing so would be desirable because the combination of Benoit, Ruihao, Nagel and Sather would be more efficient by including a quantization level of 15 when performing quantization to enhance the image quality (Bialkowski, [par. 0008]).
Response to Arguments
The examiner respectfully acknowledges the applicant’s amendments to claims 1, 3, 11, 13, and 20-22.
Applicant’s amendments filed on 03/04/2026 regarding claim objection to claims 21 and 22 have been considered and are consequently withdrawn.
Applicant’s arguments filed on 03/04/2026 regarding claim rejections to claims 1-22 under 35 USC 103 have been fully considered and moot in view of new ground of rejection (see rejections above).
Conclusion
The prior art made of record and not relied upon is considered pertinent to applicant's disclosure.
Kouri et al (US 20040071363 A1) disclosed a method for padding, filtering, denoising, image enhancing and increased time-frequency acquisition is described for digitized data of a data set is described where unknown data is estimated using real data by adding unknown data points in a manner that the padding routine can estimate the interior data set including known and unknown data to a given accuracy on the known data points.
Shi et al (US 20170270408 A1) disclosed a bit-depth optimization engine reduces the hardware cost of a neural network. When training data is applied to a neural network during training routines, accuracy cost and hardware costs are generated. A hardware complexity cost generator generates costs for weights near bit-depth steps where the number of binary bits required to represent a weight decreases, such as from 2.sup.N to 2.sup.N−1, where one less binary bit is required. Gradients are generated from costs for each weight, and weights near bit-depth steps are easily selected since they have a large gradient, while weights far away from a bit-depth step have near-zero gradients.
Milanfar et al (US 20070217713 A1) disclosed a method of creating a super-resolved grayscale image from lower-resolution images using an L.sub.1 norm data fidelity penalty term to enforce similarities between low and a high-resolution image estimates is provided. A spatial penalty term encourages sharp edges in the high-resolution image, the data fidelity penalty term is applied to space invariant point spread function, translational, affine, projective and dense motion models including fusing the lower-resolution images, to estimate a blurred higher-resolution image and then a deblurred image. The data fidelity penalty term uses the L.sub.1 norm in a likelihood fidelity term for motion estimation errors. The spatial penalty term uses bilateral-TV regularization with an image having horizontal and vertical pixel-shift terms, and a scalar weight between 0 and 1. The penalty terms create an overall cost function having steepest descent optimization applied for minimization. Direct image operator effects replace matrices for speed and efficiency.
Any inquiry concerning this communication or earlier communications from the examiner should be directed to Quoc Phung whose telephone number is (703) 756 1330. The examiner can normally be reached on Monday through Friday from 9am to 5pm PT.
Examiner interviews are available via telephone, in-person, and video conferencing using a USPTO supplied web-based collaboration tool. To schedule an interview, applicant is encouraged to use the USPTO Automated Interview Request (AIR) athttp://www.uspto.gov/interviewpractice.
If attempts to reach the examiner by telephone are unsuccessful, the examiner’s supervisor, Jennifer Welch can be reached on 571-272-7212. The fax phone number for the organization where this application or proceeding is assigned is 571-273-8300.
Information regarding the status of an application may be obtained from the Patent Application Information Retrieval (PAIR) system. Status information for published applications may be obtained from either Private PAIR or Public PAIR. Status information for unpublished applications is available through Private PAIR only. For more information about the PAIR system, see http://pair-direct.uspto.gov. Should you have questions on access to the Private PAIR system, contact the Electronic Business Center (EBC) at 866-217-9197 (toll-free). If you would like assistance from a USPTO Customer Service Representative or access to the automated information system, call 800-786-9199 (IN USA OR CANADA) or 571-272-1000.
/Q.L.P./Examiner, Art Unit 2143
/JENNIFER N WELCH/Supervisory Patent Examiner, Art Unit 2143
Prosecution Insights